July 30, 2014
Separating the Phenoms Who'll Make It From Those Who Won't
Predicting rookie players’ performances is one of the most difficult tasks for any projection system. By definition, they lack the extensive body of work necessary to analyze their performance. While it’s tempting to foresee a Hall of Fame career in the demeanor of one at-bat, statistics (as well as common sense) caution against it. No matter how locked in or lost a player looks in the sample of a few games, his true skill cannot be inferred from such meager experience.
The ultimate cautionary tale here is Mike Trout, whose first year (2011) MLB stats go like this: .220/.281/.390, in 135 plate appearances. Such a line is not the stuff from which legends are made, and had a lesser prospect posted it, there might have been rumblings of doubt in the media. Trout was decidedly not a lesser prospect, and his raw combination of discipline and athleticism had every scout and fan salivating despite the poor batting average. Even so, it’s still possible to find a few sentences in the press reports of that time which noted his difficulties:
Offensively, he has looked overmatched at times, with a .125 average (three for 24), seven strikeouts and two walks.
Granted, these utterances were often embedded in the context of articles praising Trout’s maturity, focus, or resemblance to Mickey Mantle, but my point remains: the best player in the world can flounder for his first hundred at-bats.
In the early days of a player’s career, there is a silent struggle going on between him and the rest of the league. Each side is feeling the other out, looking for weaknesses and avoiding strengths. Just ask Mike Trout, from September 2011:
It’s been tough but they are pitching to my weaknesses and I have to make adjustments…
In determining how to attack those weaknesses, opposing players have access to scouting reports and a wealth of experience. They might be the most dependable guides to a rookie’s true talent level.
Disregarding, for a moment, on-base percentage and slugging and TAv and all the other statistics in our quiver, let’s examine how the players are treating rookies and how they are responding, irrespective of their performance. The two chief tools here will be zone distance and swing distance. Zone distance defines how far from the center of each hitter’s zone batters are seeing pitches. The further away, the more respected the batter is. Swing distance measures the distance from the center of the zone for only the pitches at which a hitter swings, and so functions to describe a player’s plate discipline.
Before we look at the rookies, we should make sure that their zone and swing distances are settled down, that is to say, stabilized. Both statistics can be noisy, and so it would be desirable to find out how many pitches it takes for each to become steady. Following in the footsteps of a great kitchen implement, I looked at stabilization of zone distance.
I randomly selected a certain number of pitches (on the x-axis), calculated the median zone distance for each player, and computed the correlation (y-axis) between the sample medians and the player’s full-season stats. The crucial point—when R2 = .5—arrives at about 250 pitches worth of data. After that many pitches, the median zone distance becomes reasonably predictive of a player’s ‘true’ zone distance.
Consider the same general plot, but for swing distance.
Here, the stabilization point is reached even more quickly. At 100 swings of the bat, we have a somewhat robust estimate of the player’s plate discipline tendencies.
Reassuringly, both swing and zone distance stabilize fairly rapidly. Both statistics are derived from pitch-level measurements, and so whereas the average rookie might accrue only a few hundred plate appearances in a year, they are likely to see hundreds or even thousands of pitches. A mere 250 pitches corresponds to something like 70 PAs (at 3.5 pitches per plate appearance), at which point we ought to have some idea of how the player is being handled by the opposing pitcher. Applying the same logic to swing distance, 100 swings at a league-average swing rate (~45%) gives you about 225 pitches, or (rounding up) 65 PAs. More significant than any arbitrary cutoffs is the simple fact that both metrics stabilize rapidly.
Supposing now that these metrics are fairly consistent in small samples, let’s examine some rookies. I’ll begin with a pair of highly-touted prospects, both known for their potential future skill with the bat: Oscar Taveras and Gregory Polanco, rookies for the Cardinals and Pirates, respectively. After strong starts (Taveras hit a home run in his second at-bat), both have cooled substantially, arriving at below-average TAvs in the short time they’ve been in the majors.
Pondering Taveras first, note that his zone distance is quite solid for the year as a whole: 1.13 feet. That’s superlative, in the company of sluggers like Giancarlo Stanton and this year’s version of Justin Morneau. He couples that to a solid plate discipline profile, including a swing distance of .9 feet and a swinging strike rate of 12 percent. I suspect that Taveras is already a fine hitter, even if his performance is lagging. Cardinals’ skipper Mike Matheny has had trouble finding playing time for the young phenom, but perhaps he ought to reconsider that choice.
Polanco is less of a power threat than Taveras, and so he has a lower zone distance of 1.05. Fortunately, the man known as El Coffee is correspondingly more selective in his approach, with a swing distance of .84 feet. That’s a profile that resembles Lorenzo Cain, or Brandon Philips. The overused trope about rookies—that they struggle with breaking ball recognition—looks to be true for Polanco: he has swung at breaking pitches half an inch further from the zone, on average, than fastballs. But, barring that one weakness, the zone distance plot shows that Polanco seems to be a steadily average hitter, a lofty achievement for a 22-year old who plays good defense.
Young Xander Bogaerts acquitted himself more than adequately in last year’s championship squad, but is now struggling mightily. Last year’s TAv mark of an even .300 has sunk to .239. His current zone distance of .98 gives him a bottom 10 grade, which is not ideal. Gazing at his zone distance trend shows some positive development, however.
Even though Bogaert’s full-season stats are disappointing, his zone distance trend is positive, if levelling off. Depending on whether he’s able to elevate it some more in the latter half of the season, Bogaerts might be a sneaky breakout candidate. At minimum, he’s likely to overperform his stats this year.
Not all rookies are excelling. I have bad news for New York Mets fans: Travis d’Arnaud isn’t scaring anyone. His current, full year zone distance of .96 feet would rank last among hitters with more than 1000 pitches seen so far this season. In and of itself, starting off with a low zone distance wouldn’t be so bad, but it is more troubling to see what d’Arnaud has been able to do with all of these pitches at the center of the zone: a .223/.290/.350 line. It’s natural for pitchers to try to probe a young player’s weaknesses, but no hitter’s weakness should include the middle of the plate. Looking at his trend, there’s no obvious tendency to force pitchers away from the zone.
Nothing is set in stone for any rookie (or indeed, any player). As described above, the early appearances of a hitter are all about adjustments and counter-adjustments. A skillful slugger is able to compensate for the opposing pitcher’s strategy. So, while the distance metrics may stabilize early on for all hitters, they are probably a little more volatile for younger hitters.
Here’s a mystery player. In his first few hundred plate appearances, he managed a decent but unspectacular 1.03 feet of zone distance, with a moderate swing distance of .89 feet. Like Gregory Polanco, he struggled with breaking ball recognition, but to an even worse extent—he swung at breaking balls two inches further away, on average, than fastballs. He didn’t manage great stats, but he was young and impressed his teammates nonetheless.
The mystery player in question is Mike Trout, circa 2011. Before he became world-beating and historically unprecedented, he was bamboozled by curves and sliders. All of which is to say that rookies can (and do) change, improve, and sometimes defy expectations. It’s easy to lose sight of what made a superprospect so exciting when they look overmatched at the plate, but MLB has a learning curve, and a distance-based approach shows that things are by and large not as dire as they at first appear*. Many rookies struggle to make their slash line match their abilities, but to the extent that the viewpoints of opposing pitchers are valuable, much of the current crop of high-profile talent is likely in a position to succeed.
*Except for Travis d’Arnaud.